Summary of A Bayesian Approach to Harnessing the Power Of Llms in Authorship Attribution, by Zhengmian Hu et al.
A Bayesian Approach to Harnessing the Power of LLMs in Authorship Attribution
by Zhengmian Hu, Tong Zheng, Heng Huang
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Applications (stat.AP)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Medium Difficulty summary: This study investigates the potential of Large Language Models (LLMs) in one-shot authorship attribution, a task that aims to identify the origin or author of a document. Traditional approaches have limitations due to manual features and lack of capturing long-range correlations. Recent advancements leveraging pre-trained language models require fine-tuning on labeled data, posing challenges in data dependency and limited interpretability. The authors utilize Bayesian approaches and probability outputs of LLMs, such as Llama-3-70B, to calculate the probability that a text entails previous writings of an author. This methodology reflects a more nuanced understanding of authorship. The results show impressive 85% accuracy on IMDb and blog datasets across ten authors in one-shot authorship classification. This work sets new baselines for one-shot authorship analysis using LLMs, expanding their application scope in forensic linguistics. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Low Difficulty summary: This research looks at how to identify the author of a document just by reading it once. Current methods have limitations because they rely too much on human judgment and don’t capture long-term connections between words. The authors use special language models that can reason deeply and understand complex relationships between words. They test these models on movie reviews and blogs from ten different authors, achieving an impressive 85% accuracy in just one read-through. This work opens up new possibilities for using these language models to analyze authorship in fields like forensic linguistics. |
Keywords
» Artificial intelligence » Classification » Fine tuning » Llama » One shot » Probability
